boardroom mcp shipped on hn today - multi-advisor governance engine for ai agents. solo founder pattern, careful design, the project deserves the karma it picked up.
there's one architectural detail that decides whether multi-advisor governance is a demo or a procurement-ready product - the 'who said what when' log.
what most multi-agent coordinators ship
- agent A votes on action X
- agent B votes on action X
- coordinator picks the majority
- log says 'action X executed'
what an audit needs 14 months later
- agent A's vote, signed, with the prompt that produced it
- agent B's vote, signed, with the prompt that produced it
- coordinator's tie-break logic, version-pinned
- merkle inclusion proof showing all three were captured at the same instant
- the operator authorization that enabled the run
why this matters for the audit
in a regulated industry, 'the multi-agent system decided' is not a defensible answer. the auditor will ask which advisor agent said what, with what input, at what timestamp, signed by whose key. if the coordinator can't reconstruct that 14 months later, the entire governance chain is hearsay.
the bizsuite layer
await bizsuite.audit.advisor_vote({
council_id, advisor_id, vote, prompt_hash,
signature, ts
});
await bizsuite.audit.coordinator_decision({
council_id, decision, votes_seen,
tie_break_logic_version, ts
});
two calls. hash-chained. exportable.
why solo-founder governance tools matter
the enterprise governance vendors are 18 months behind on multi-agent patterns. the solo-founder ones are shipping the right architecture. the gap they have is the procurement-ready report. bizsuite ships that. $997.
Top comments (0)